Published on October 15, 2007
Topology Measurements: Topology Measurements Subprogram 1, Workpackage 1a (DIMES) Participating partners: Participating partners Tel Aviv University Collegium Budapest The Hebrew University Goals 2006-2007: Goals 2006-2007 To position DIMES as a leading infrastructure for Internet topology research. To increase system openness for data and experiments. To stabilize the system engineering. To develope permanent interoperability between ETOMIC and DIMES. To study the Internet topology To study the Internet evolution in time and build a model. Deliverables: Deliverables 36 Months: D.1a.3 Web-based interface for query access to live DIMES databases . 48 Months: D.1b.4 A Model of Internet evolution. Milestones: Milestones M.1b-3.1 Implementation of BART tool in ETOMIC [M32]. M.1b-3.2 ETOMIC-DIMES permanent interoperability [M32]. M.1b- 3.3 Final ETOMIC visualization tools [M36]. M.1b-3.4 European periodic traffic tomography map [M36]. M.1b-3.5 Final form and documentation of BART [M48]. M.1b-3.6 Advanced final ETOMIC measurement archives [M48]. Meetings in 2006: Meetings in 2006 February: ETOMIC workshop in Budapest Participants: Colbud, TAU, Pamplona, Lousanne, Ericsson, HUJI -- Agreement on the joint implementation steps of the ETOMIC-DIMES measurements -- Agreement on the joint implementation steps of BART in ETOMIC -- Agreement on the steps of the WP1c whitepaper May: Meeting with the MoMe (IST Cluster Project) May: Ohad Serfaty (TAU) and Jozsef Steger (Colbud) work in Budapest and integrate the pattern generator into DIMES September: Meeting in Budapest with Darlagiannis Vasilios ( Lausanne) to finalize the white paper December: Rome 3rd Year Achievements: 3rd Year Achievements A basic packettrain module (with ColBud) Integrated in the current agent Parameters: packt size, train length, protocol A new packettrain module (with ColBud) Enhancements to measure QoS support Tested against ETOMIC cards Compiled into the agent Initial results are now analyzed Web Experiment Planner Allow anyone to run its own DIMES experiment Supports also XML uploads Periodic maps on the web Topologies: AS, router, metro In a few weeks: delay maps In a few month: PoP level maps Research areas: Research areas k shell analysis (with HUJI) Network algorithmica Shortest paths Generalized Efficient median calculation Constant space, linear time Automatic PoP idetification Based on traceroute information No need for router identification Routing stability IRR validity 4th Year: 4th Year Visualization 4D graph visualization (done) Database visualization (almost done) Routing visualization (by year end) Distance service Will be enhanced Estimation algorithm for large maps (100,000s of nodes) – by summer 2007 Integration with eMule (done) 15-20% improvement in throughput 4th Year: 4th Year DIMES- ETOMIC experiments to continue on wide scale Data service (initial concept finished, be done in 3 month) DIMES on mobile (with HUJI) s/w demo by HUJI Contact with Motorola for porting P2P – large scale query study on Gnutella Static/dynamic/geographic analysis First results based on 665M queries in 4 months (38 days) were published. Continue research on profiling, media event success. DIMES Status: DIMES Status Over 5400 users, over 12,000 agents 94 countries All continents Over 700 ASes More than 1500 are active daily Over 5,000,000 measurements a day Data is used world wide Experiment Planning: Experiment Planning DIMES Visualizer: DIMES Visualizer DIMES Internet Mapper: DIMES Internet Mapper Median Estimation: Median Estimation The algorithm uses 1 path (streaming algorithm) and two variables for the estimation. Better than any known algorithm. Gives indication on the results accuracy. Median Estimation: Median Estimation Delay estimation Vs. propagation delay lower bound Estimation behaves as a normal variable Thank you!: Thank you!